Neural network based adaptive out-of-step protection strategy for electrical power systems

Application of artificial intelligence to power systems has resulted in an overall improvement of solutions in many areas. This paper presents a new strategy for adaptive out-of-step protection of synchronous generators based on neural networks. The paper describes the neural networks architecture adopted as well as the selection of input features for training the neural networks. A feed forward model of the neural network based on the stochastic back-propagation training algorithm has been used to predict the out-of-step condition. Due to power network configuration changes, the performance of the protective relays can vary. Consequently, a new adaptive out-of-step protection strategy is suggested in this paper. This depends firstly on detecting the case of the system through case detection neural networks by some pre-fault local measurements at the machine to be protected, and then calculating the new out-of-step condition through an adaptive routine. The capabilities of the developed adaptive out-of-step prediction algorithm have been tested through computer simulation for a typical case study. The results of using the proposed algorithm demonstrate the adaptability of the proposed strategy.